Deep Learning Approaches for Marine Oil Spill Detection and Monitoring

Abstract

Oil spill catastrophes critically affect marine environments and coastal economies, resulting in long-term environmental and economic damage. Early identification is essential to reduce damage, but conventional approaches based on Convolutional Neural Networks (CNNs) are encumbered with low accuracy and scalability when working with large datasets for real-time observation. These concerns necessitate improved solutions to enhance efficiency and trustworthiness in oil spill detection. For these problems, the YOLO v8 model has been suggested for oil spill detection. YOLO is a cutting-edge real-time object detection algorithm that outperforms conventional CNN-based approaches in speed and accuracy. YOLO v8 improves upon earlier versions with better detection accuracy and efficient processing of large datasets. Its efficient architecture supports single-pass image analysis, ensuring quick detection of oil spills. This makes it especially ideal for time-critical applications where quick responses are critical to reduce environmental and economic impacts. The inclusion of YOLO v8 in detection systems allows for constant monitoring using satellite or drone imagery, greatly enhancing detection speed and scalability for large water bodies. This technology advances environmental protection and enhances disaster response operations, making it a revolutionary tool in mitigating the impacts of oil spill catastrophes.